Computational Intelligence Methods Research Articles

Sensitivity Analysis of Sensors in a Hydraulic Condition Monitoring System Using CNN Models.

Condition monitoring (CM) is a useful application in industry 4.0, where the machine’s health is controlled by computational intelligence methods. Data-driven models, especially from the field of deep learning, are efficient solutions for the analysis of time series sensor data due to their ability to recognize patterns in high dimensional data and to track the temporal evolution of the signal. Despite the excellent performance of deep learning models in many applications, additional requirements regarding the interpretability of machine learning models are getting relevant. In this work, we present a study on the sensitivity of sensors in a deep learning based CM system providing high-level information about the relevance of the sensors. Several convolutional neural networks (CNN) have been constructed from a multisensory dataset for the prediction of different degradation states in a hydraulic system. An attribution analysis of the input features provided insights about the contribution of each sensor in the prediction of the classifier. Relevant sensors were identified, and CNN models built on the selected sensors resulted equal in prediction quality to the original models. The information about the relevance of sensors is useful for the system’s design to decide timely on the required sensors.

Analytical Modelling of Share Price Value Using Computational Intelligence Methods

The liberalization, volatility and high competition of financial markets are all factors that expose companies to new risks and challenges, requiring a continuous innovation of models, techniques and tools for managing share price value and other related risks. This paper provides a model to implement a subsystem that should support the management decision on the trading segment of its own shares, based on intelligent agents with regards to identifying, collecting, structuring and updating data instruments, with analysis mechanisms of the main price and volatility indicators. Our conclusion is that the use of computational intelligence methods in modeling of share price value is both a requirement and an option in managing financial-foreign exchange risks, given that companies are subject to a wide range of risks and volatility is the defining feature in which they operate.

Classification of Electrical Power System Conditions with Convolutional Neural Networks

Undesirable operation of a distant relay at the occurrence of stressed conditions is a reason for blackouts. There are a few computational intelligent methods available in the literature for avoiding relay maloperations. However, because of variations in the system parameters and expansions of the network, the performance of these techniques can be degraded. To solve this issue, data mining approaches have been introduced. The existing data mining approaches need improvement in terms of accuracy and error rate while discriminating fault and stressed conditions. In this paper, a Convolutional Neural Network (CNN) based classifier is proposed for identifying various faults and differentiating fault and power swing situations. The data are collected from the IEEE-9 bus system by Phasor Measurement Units (PMU) and the proposed CNN classifier model classifies system conditions like normal, fault, and power swing. The outcome shows that the classifier has high accuracy and low error rate compared to other classification models such as Naïve Bayes, Decision Tree, and K-Nearest Neighbor. Furthermore, the proposed CNN model is validated with the use of TensorFlow framework to demonstrate its superior performance.

Expression of Concern: A hybrid computational intelligence method for predicting dew point temperature

The Editors-in-Chief of Environmental Earth Sciences are issuing an editorial expression of concern to alert readers that this article.

Construction of the Operating Limits Diagram for a Ship-Based Helicopter Using the Design of Experiments with Computational Intelligence Techniques

Compared with land-based helicopters, ship-based helicopters are required to land in a more challenging working environment as the airwakes generated by the wind field flowing through the superstructure of the ship changes the wind field structure. This complicates the wind field structure and affects the safety of flight control. The flight safety of the helicopter pilot can be significantly improved with prior understanding of the relevant information in the ship-based helicopter operating limits (SHOL) diagram. In previous studies, the SHOL diagram of ship-based helicopters has been obtained using numerical simulations in conjunction with a flight simulator. However, the flight simulator equipment is expensive and difficult to maintain. This study references the aforementioned studies by initially employing a numerical simulation method to obtain the flow field information of the interaction between the airwakes of the ship’s superstructure and the downwash flow of the helicopter. Then, the flight simulator is replaced by computational intelligence methods involving artificial intelligence. This significantly reduces the research cost for envelope construction. This study integrates design of experiments (DOE) and computational intelligence techniques (soft computing) to establish a recommended range for the SHOL diagram of ship-based helicopters. This study utilizes the DOE and computational intelligence techniques to construct the SHOL diagram of ship-based helicopters, provide suggestions, and serve as a reference for helicopter pilots and engineering designers to improve the safety during flight.

Prostat MRT tərəfindən prostat xərçəngi xəstələrində patoloji mərhələnin proqnozlaşdırılması: Süni sinir şəbəkələri metodları

Prostate cancer is a disease that is most common in males and causes death in the second frequency in the world. If prostate cancer is diagnosed in the early stages, surgery can be performed and the disease can be cured. The aim of this study is to design an expert system to catch prostate cancer as early as possible with the chance of surgical treatment by being diagnosed in the limited phase of the organ. The most accurate diagnosis is to use risk factors such as Age, PSA (prostate Specific antigen), Clinical Stage, Tumor Size, Prostate Size and ISUP (International Society of Urological Pathology). In other words, it is aimed to biopsy the minimum number of patients and to diagnose the largest number of cancers. For better detection both sets of characteristics are used in our research. In this study, as a diagnostic model, we use a system based on multiple-layer (deep) feed-forward neural networks. The networks are trained with Differential Evolution training algorithm using in parallel a pair of data sets (training and validation sets) to avoid overfitting and improve model’s generalization ability (performance on untrained data). The applied DE algorithm has allowed avoiding local minima of error function during the training. A third data set is used for testing trained network performance. According to the obtained results, this method demonstrated better results than other existing approaches. Key words: Prostate Cancer, Artificial Neural Network, Differential Evolution Optimization Computational Intelligence Methods

Application of Computational Intelligence in Describing the Drying Kinetics of Persimmon Fruit (Diospyros kaki) During Vacuum and Hot Air Drying Process

This study examines the potential of applying computational intelligence modelling to describe the drying kinetics of persimmon fruit slices during vacuum drying (VD) and hot-air-drying (HAD) under different drying temperatures of 50 °C, 60 °C and 70 °C and samples thicknesses of 5 mm and 8 mm. Kinetic models were developed using selected thin layer models and computational intelligence methods including multi-layer feed-forward artificial neural network (ANN), support vector machine (SVM) and k-nearest neighbors (kNN). The statistical indicators of the coefficient of determination (R2) and root mean square error (RMSE) were used to evaluate the suitability of the models. The effective moisture diffusivity and activation energy varied between 1.417 × 10−9 m2/s and 1.925 × 10−8 m2/s and 34.1560 kJ/mol to 64.2895 kJ/mol, respectively. The thin-layer models illustrated that page and logarithmic model can adequately describe the drying kinetics of persimmon sliced samples with R2 values (>0.9900) and lowest RMSE (<0.0200). The ANN, SVM and kNN models showed R2 and RMSE values of 0.9994, 1.0000, 0.9327, 0.0124, 0.0004 and 0.1271, respectively. The validation results indicated good agreement between the predicted values obtained from the computational intelligence methods and the experimental moisture ratio data. Based on the study results, computational intelligence methods can reliably be used to describe the drying kinetics of persimmon fruit.

An extensive review of computational intelligence-based optimization algorithms: trends and applications

Area of computational intelligence is gaining researcher’s attention in ongoing trend of technology and evolution due to their high capability to deliver near-optimal solutions. A new hierarchy of algorithms has been proposed in the paper, and they have been organized on the basis of their inspiration sources. The broad two domains of the algorithms are modeling of human mind and nature-inspired intelligence. Nature-inspired computational algorithms being heuristic algorithms are robust and have optimization capability to solve obscure and substantiated problems. The heuristic techniques aim on finding the best possible solution to the query in a satisfiable amount of time. The computational intelligence methods inspired from nature have further been categorized into artificial immune systems, evolutionary algorithms, swarm intelligence, artificial neural networks and geoscience-based algorithms. Geoscience-based domain is the least explored domain in which the algorithms can be developed based on geographic phenomenon taking place on the earth’s surface. An extensive tabular comparison is done among algorithms of all the domains on the basis of various attributes. Also, variants of the algorithms and their implementation in a specific application have been examined. The efficiency and performance of selected algorithms have been compared on clustering and traveling salesman problem for better understanding.

A new framework for predicting customer behavior in terms of RFM by considering the temporal aspect based on time series techniques

Data mining and in particular forecasting tools and techniques are being increasingly exploited by businesses to predict customer behavior and to formulate effective marketing programs. Conventionally, customer segmentation approaches are utilized when dealing with a large population of customers. Inspired by this idea, a new methodology is proposed in this study to perform segment-level customer behavior forecasting. To keep the dynamic nature of customer behavior, customer behavior is represented as a time series. Therefore, customer behavior forecasting is changed into a time series forecasting problem. The proposed methodology contains two main components i.e. clustering and forecasting. In the clustering phase, time series are clustered using time series clustering algorithms, and then, in the forecasting phase, the behavior of each segment is predicted via time series forecasting techniques. The main objective is to predict future behavior at segment level. The forecasting component also consists of a combined method exploiting the concept of forecast fusion. The combined method employs a pool of forecasters both from traditional time series forecasting and computational intelligence methods. To test the usefulness of the proposed method, a case study is carried out using the data of customers’ point of sale (POS) in a bank. The results of the experiments demonstrate that the combined method outperforms all other individual forecasters in terms of symmetric mean absolute percentage error (SMAPE). The proposed methodology can be correspondingly applied in other areas and applications of time series forecasting.

Coverage and k-Coverage Optimization in Wireless Sensor Networks Using Computational Intelligence Methods: A Comparative Study

The domain of wireless sensor networks is considered to be among the most significant scientific regions thanks to the numerous benefits that their usage provides. The optimization of the performance of wireless sensor networks in terms of area coverage is a critical issue for the successful operation of every wireless sensor network. This article pursues the maximization of area coverage and area k-coverage by using computational intelligence algorithms, i.e., a genetic algorithm and a particle swarm optimization algorithm. Their performance was evaluated via comparative simulation tests, made not only against each other but also against two other well-known algorithms. This appraisal was made using statistical testing. The test results, that proved the efficacy of the algorithms proposed, were analyzed and concluding remarks were drawn.

Improving the reliability of test functions generators

Computational intelligence methods have gained importance in several real-world domains such as process optimization, system identification, data mining, or statistical quality control. Tools are missing, which determine the performance of computational intelligence methods in these application domains in an objective manner. Statistics provide methods for comparing algorithms on certain data sets. In the past, several test suites were presented and considered as state of the art. However, there are several drawbacks of these test suites, namely: (i) problem instances are somehow artificial and have no direct link to real-world settings; (ii) since there is a fixed number of test instances, algorithms can be fitted or tuned to this specific and very limited set of test functions; (iii) statistical tools for comparisons of several algorithms on several test problem instances are relatively complex and not easily to analyze. We propose a methodology to overcome these difficulties. It is based on standard ideas from statistics: analysis of variance and its extension to mixed models. This paper combines essential ideas from two approaches: problem generation and statistical analysis of computer experiments.

Shallow Landslide Susceptibility Mapping by Random Forest Base Classifier and Its Ensembles in a Semi-Arid Region of Iran

We generated high-quality shallow landslide susceptibility maps for Bijar County, Kurdistan Province, Iran, using Random Forest (RAF), an ensemble computational intelligence method and three meta classifiers—Bagging (BA, BA-RAF), Random Subspace (RS, RS-RAF), and Rotation Forest (RF, RF-RAF). Modeling and validation were done on 111 shallow landslide locations using 20 conditioning factors tested by the Information Gain Ratio (IGR) technique. We assessed model performance with statistically based indexes, including sensitivity, specificity, accuracy, kappa, root mean square error (RMSE), and area under the receiver operatic characteristic curve (AUC). All four machine learning models that we tested yielded excellent goodness-of-fit and prediction accuracy, but the RF-RAF ensemble model (AUC = 0.936) outperformed the BA-RAF, RS-RAF (AUC = 0.907), and RAF (AUC = 0.812) models. The results also show that the Random Forest model significantly improved the predictive capability of the RAF-based classifier and, therefore, can be considered as a useful and an effective tool in regional shallow landslide susceptibility mapping.

Computational Intelligence Techniques Used for Stock Market Prediction: A Systematic Review

With the advancement of various computational techniques and the growing search for assertive predictive models, computational intelligence methods have attracted much attention. They are data-based methodologies and mainly include fuzzy logic, artificial neural networks and evolutionary computation. In the economic environment, more specifically, in the stock market forecast, where there is the challenge of the time series volatility, these methods have stood out. In this context, the objective of this paper is to present a systematic review of the literature on recent research involving forecasting techniques in the stock market, and the computational intelligence were the ones that stood out. To define these techniques, articles were collected from four large databases and a keyword filter was applied, which reduced the initial volume. So we selected the articles from the most published journals and remove duplicated articles. The most articles applied hybrid models and for the selection of featured techniques were choose those most frequent ones. A brief description was also made of the most used methods as well as of the selected articles. The review was done with articles published between the years 2014 and 2018 taken from four databases and, after some selection criteria, 24 articles were selected by relation to the subject studied.

Determining Electoral Preferences in Mexican Voters by Computational Intelligence Algorithms

In the context of political activities, electoral processes are of interest for scientists, who usually tackle their research on this field from a social sciences perspective. Computational methods have been applied to predict the electoral preferences of voters in several countries; however, this has not happened in Mexico, at least as indicated by the absence in current scientific literature of computational studies to determine voting intentions of Mexican citizens. The authors of the present work aim at reverting such absence. The proposal of this paper consists of applying Computational Intelligence methods to automatically determine electoral preferences of Mexican voters. For this, data acquired by the Secretaría de Gobernación (Secretary of the Interior), about voting intentions of Mexican citizens in the 2012 elections are used. In the voter classification stage, a modified version of the Gamma Associative Classifier (MGAC) is used, given that this is one of the relevant models of the Associative approach to Pattern Classification. Additionally, Differential Evolution is employed to guide the process of relevant features selection. Results indicate that, when compared over six data sets extracted from the information published by the Secretaría de Gobernación, our proposal exhibits the best performance in three of these data sets, outperforming some of the best similar models present in the state of the art.

An optimized model using LSTM network for demand forecasting

In a business environment with strict competition among firms, accurate demand forecasting is not straightforward. In this paper, a forecasting method is proposed, which has a strong capability of predicting highly fluctuating demand data. Therefore, in this paper we propose a demand forecasting method based on multi-layer LSTM networks. The proposed method automatically selects the best forecasting model by considering different combinations of LSTM hyperparameters for a given time series using the grid search method. It has the ability to capture nonlinear patterns in time series data, while considering the inherent characteristics of non-stationary time series data. The proposed method is compared with some well-known time series forecasting techniques from both statistical and computational intelligence methods using demand data of a furniture company. These methods include autoregressive integrated moving average (ARIMA), exponential smoothing (ETS), artificial neural network (ANN), K-nearest neighbors (KNN), recurrent neural network (RNN), support vector machines (SVM) and single layer LSTM. The experimental results indicate that the proposed method is superior among the tested methods in terms of performance measures.

Computationally intelligent modelling of the plasma cutting process

ABSTRACTThis paper investigates the applicability of two computational intelligence methods for the plasma cutting process modelling: artificial neural networks and adaptive neuro-fuzzy inference systems. After exploring the possibilities of neural networks learning from the experimental data for the prediction of the plasma cutting parameter, the adaptive neuro-fuzzy inference system approach was used in order to compare the prediction properties of neural networks and adaptive neuro-fuzzy models. These two methods resulted in the kerf width prediction models for different combinations of input process parameters: cutting current, cutting speed and material thickness, whose significance was assessed by multi-criteria analysis. Descriptive statistics and a visual exploration of two sets of neural-network and adaptive neuro-fuzzy modelled data showed good agreement with the experimental results and their trends. This was additionally confirmed by the t-test and Analysis of Variance, which also provided the selection of the most favourable method for plasma cutting modelling. Accordingly, the adaptive neuro-fuzzy inference systems showed superior modelling capabilities over artificial neural networks for this particular problem setting.

Predição de tendências em séries financeiras utilizando metaclassificadores

Neste trabalho foi desenvolvido um metaclassificador baseado em métodos de inteligência computacional para prever tendências em séries temporais financeiras. O kernel do metaclassificador foi baseado na ferramenta (Weka). Sete classificadores foram combinados para realizar a metaclassificação. Testes foram realizados com nove ativos da Bolsa de Valores de São Paulo. Os resultados iniciais foram promissores, com boa acurácia na classificação e ganhos de até 100% do valor de capital inicialmente investido no período de um ano.

Hybrid Computational Intelligence Methods for Landslide Susceptibility Mapping

In this study, hybrid integration of MultiBoosting based on two artificial intelligence methods (the radial basis function network (RBFN) and credal decision tree (CDT) models) and geographic information systems (GIS) were used to establish landslide susceptibility maps, which were used to evaluate landslide susceptibility in Nanchuan County, China. First, the landslide inventory map was generated based on previous research results combined with GIS and aerial photos. Then, 298 landslides were identified, and the established dataset was divided into a training dataset (70%, 209 landslides) and a validation dataset (30%, 89 landslides) with ensured randomness, fairness, and symmetry of data segmentation. Sixteen landslide conditioning factors (altitude, profile curvature, plan curvature, slope aspect, slope angle, stream power index (SPI), topographical wetness index (TWI), sediment transport index (STI), distance to rivers, distance to roads, distance to faults, rainfall, NDVI, soil, land use, and lithology) were identified in the study area. Subsequently, the CDT, RBFN, and their ensembles with MultiBoosting (MCDT and MRBFN) were used in ArcGIS to generate the landslide susceptibility maps. The performances of the four landslide susceptibility maps were compared and verified based on the area under the curve (AUC). Finally, the verification results of the AUC evaluation show that the landslide susceptibility mapping generated by the MCDT model had the best performance.

An adaptive ensemble classification framework for real-time data streams by distributed control systems

Smart Grids are critical infrastructure networks. They play a critical role in the survival of our postmodern economies, as all other areas depend on their availability. An interruption in their operation may have a direct impact on the availability of other services (e.g., health, transportation). The problem is particularly intense when no backup networks are available, or the required recovery time is beyond backup autonomy. The transition to a decentralized management and control system for these networks requires digital technologies, advanced interconnected system communications, and Internet access. These technologies expose critical infrastructure networks to external threats that require careful assessment of cyber-security risks and appropriate countermeasures. An important factor that enhances the range of threats is the heterogeneity of Smart Grids, which incorporate industrial control systems such as the SCADA, distributed control system, and programmable logic controllers to which security improvements may not have been made since they were installed. Υet, another serious problem arises from the fact that older technologies were designed at times when cyber-security was not part of their technical design specifications. At the same time, it should be seriously considered that many of the systems of these networks that can be cyber-attacked may not be easily disconnected, as this could potentially lead to generalized operational problems. In this scientific research, a sophisticated active security framework is proposed, which is based solely on advanced computational intelligence methods and concerns the digital security of critical infrastructure networks. Specifically, this research introduces a sophisticated adaptive ensemble classification framework for real-time data streams by distributed control systems. It is a “Kappa” architecture framework that is based on a two-step online ensemble learning model based on bagging and boosting methods. The aim is performance of real-time analysis and evaluation of data flows from Smart Grids, toward the effective identification of APT attacks.

Stock Market Forecasting Using Computational Intelligence: A Survey

Stock market plays a key role in economical and social organization of a country. Stock market forecasting is highly demanding and most challenging task for investors, professional analyst and researchers in the financial market due to highly noisy, nonparametric, volatile, complex, non-linear, dynamic and chaotic nature of stock price time series. Prediction of stock market is a crucial task and prominent research area in financial domain as investing in stock market involves higher risk. However with the development of computational intelligent methods it is possible to reduce most of the risk. In this survey paper, our focus is on application of computational intelligent approaches such as artificial neural network, fuzzy logic, genetic algorithms and other evolutionary techniques for stock market forecasting. This paper presents an up-to-date survey of existing literature on stock market forecasting based on computational intelligent methods. In this article, the selected papers are organized and discussed according to six main point of view: (1) the stock market analyzed and the related dataset, (2) the type of input variables investigated, (3) the pre-processing techniques used, (4) the feature selection techniques to choose effective variables, (5) the forecasting models to deal with the stock price forecasting problem and (6) performance metrics utilized to evaluate the models. The major contribution of this work is to provide the researcher and financial analyst a systematic approach for development of intelligent methodology to forecast stock market. This paper also presents the outlines of proposed work with the aim to enhance the performance of existing techniques.